Last season, we tried using the free ADXL362/ADXRS450 breakout board from FIRST Choice for autonomous mode positioning. We never really got far on it, but one of the biggest issues we saw was large sensor noise from the accelerometer. I expect some noise since robots aren’t smooth, but we were getting constant +/-1g noise with no real noise frequency that we could find to filter.
A lot of that noise was probably self induced:
Wheels were not “round” (Vex 10" wheels with the thread not fully smooth)
Control board was plywood not firmly mounted to chassis (tie wraps to bottom support boards, and RoboRio mounted with foam tape on board with tie wraps to help hold)
Breakout board plugged directly into RoboRio (I’m sure the other end of the board had some give)
So my question is, are there any suggestions on how we can mount the sensor better? I am assuming that we should move the sensor with a ribbon extension cable, but any other suggestions? I find mounting it in silly putty (https://www.chiefdelphi.com/forums/showpost.php?p=1114729&postcount=15) as a cool idea, anyone else tried this? Any pictures of it? Did you worry about it conducting on the putty? (I heard kneaded gray erasers work too)
I attached some of the data we recorded from the robot as examples. This was collected by saving the NetworkTables buffer internally in the RoboRio.
This probably didn’t do your control wiring any good, either. We had our STRONHOLD control board mounted with rubber pads and bungee, which allowed slight movement and some shock absorption, better both for controls and gyros.
While it isn’t really critical at the FRC time and size scales, we still usually try to mount our gyros/inertial navigators to the chassis near the center of gravity of the robot. It should not be directly screwed or tied down, but there should be some fairly stiff shock mounting, similar to how you’d mount a compressor. For us, this has always meant that it was a good bit away from the RIO. IIRC, our last gyro was held down with a dense foam tape, with a cage of tie-wraps to capture it in case it came unstuck.
971’s was solidly bolted to the belly pan in order to couple the high frequency signals to the gyro, and worked great. I’d recommend coupling the gyro and accelerometer to the robot rigidly to avoid modifying the frequency content. You are better off filtering it in software.
We ran the ADXRS453 last year to very good effect.
I’ve never had good luck with accelerometers. The noise (and the second integral of it) is killer, as I think you are showing in your data. What are you trying to do with one that an encoder couldn’t do?